GNGTS 2019 - Atti del 38° Convegno Nazionale

GNGTS 2019 S essione 2.2 351 The completely different polarization between the broad peak (ranging from 6 Hz and 10 Hz) and the occasional peak (at 3.3 Hz), respectively 160 degree and 60 degree, is another very interesting feature of this site. Discussion. For the seven analyzed sites the HVSR curve obtained from the analysis of seismic noise is not a stable feature. In fact the typical shape of the HVSR curve obtained for each site in case of strong noise differs quite a lot from that obtained in case of weak noise. Five sites are located on hill slope (MMN and SMIN) or ridge crest (TP02, ROI, CMPN), in places where the soft soil thickness can be neglected. In the HVSR curves of these sites computed on weak noise we expect contributions from both geology and topography (Burjanek et al. , 2014), not easily discernible from each other. For all these five sites the occasional contribution to the HVSR yields peaks in a broad frequency range, from 1.7 Hz to 14 Hz. The two sites at Tortora Marina, T002 and T017, located in between the sea and the mountain, show always a well defined persistent peak at 2.15 Hz and 1.4 Hz, respectively. The occasional peak that appear in case of strong noise have lower frequency, 1.3 Hz and 0.95 Hz. We believe that at these two sites the occasional peak is likely the result of a shear head wave produced by the sea waves as they approach the coast. The HVSR peak associated with shear head waves has been theoretically described by Bonnefoy-Claudet et al. (2006a). Unfortunately, our single station data are not sufficient to confirm or reject this hypothesis. All the persistent HVSR peaks observed in the seven sites are well polarized in the horizontal plane. In mountain environments where topographic effects occur, the polarization is parallel to slope direction (MMN, SMIN) or normal to ridge crest (TP02, ROI, SMIN), while along the coast (T002, T017) the polarization of the peaks, due to resonance of soft layer, is parallel to the presumed layer dipping direction. All analyzed data were acquired by seismic stations installed inside a building (house basement, cellar, vault, and similar locations) with the seismometer covered by a shelter and thermally insulated in case of broad band sensor, therefore the direct action of any external factors upon the instruments (wind, rain etc..) cannot be responsible of the observed HVSR variations. We suppose that the composition of the seismic wavefield has a greater amount of body waves in case of strong noise compared to weak noise. The interaction of body waves with local geology and topography is expected to produce different effects compared with the interaction of surface waves. The contents of surface and body waves in the background signal are not constant and depends on the noise sources, their features and distance from the recording site. In the hypothesis that strong noise has a contents of body waves greater than weak noise, we can explain the occasional low frequency peak that rise up in the HVSR at the two sites near the coast, T002 and T017, as produced by shear head waves, as suggested by some theoretical analyses (Bonnefoy-Claudet et al. , 2006a). Furthermore, the interaction of body waves with the local topography could explain the different HVSR curves observed for sites located on hill slope or upon a ridge. The important contribution of body waves to topographic effects has been recognized in many studies (e.g. Chavez-Garcia et al. , 1996), therefore we believe it is the most likely source of the occasional effects observed during strong noise periods. Conclusions. While the stability of HVSR results in simple geological contexts is well supported by many papers, our study demonstrates that in some cases the HVSR results are not stable in time and their interpretation is not obvious, especially at sites characterized by rough topography and complex geological structure. The results of HVSR analysis computed for different time periods may be very different, with a tight relationship with the amplitude of the incoming waves. Understanding how the composition of the seismic signals, in particular the amount of body waves in the background noise, affects the HVSR is fundamental in places where different site effects may occur. However, a better comprehension of the seismic wavefield, for example from the analysis of array data recordings, and a good knowledge of the local geological structure which include a 3D velocity model, are necessary to explain the occasional HVSR peaks.